Telexed

AI news through a solo-operator lens — only what changes your day5 of 829

FILTER[All][Agents & tools][Models & API][Generative media][Infra & SaaS][ASO & growth][Indie business][Idea signals][Other][★6+ high-signal]

Claude Code Releases ✕clear filters

Yesterday2 dispatches

#0829
#0829Agents & tools Claude Code Releases13 hours ago
`Claude Code` `v2.1.158` adds Auto mode on Bedrock, Vertex, and Foundry
60radar
Claude CodeCoding agent CLI — automates code work with Claude
Auto mode now reaches managed-cloud deployments for Opus 4.7 and Opus 4.8. Useful only if you run through Bedrock, Vertex, or Foundry.
- CLAUDE_CODE_ENABLE_AUTO_MODE=1 enables the feature, so rollout is opt-in rather than default behavior.
- Support covers Bedrock, Vertex, and Foundry for Opus 4.7 and Opus 4.8; direct API users get no new signal here.
- Practical value is higher for teams keeping LLM traffic inside cloud-vendor rails; local CLI-only workflows can skip this update.
Source: github.com/anthropics/claude-code/releases/tag/v2.1.158Read original →
FIG-8291:1
60radar
FIG-8291:1
#0828
#0828Agents & tools Claude Code Releases19 hours ago
`Claude Code` `v2.1.157` auto-loads local skill plugins
80radar
Claude CodeCoding agent CLI — automates code work with Claude in terminal
Local .claude/skills plugins now load without marketplace setup. Plugin scaffolding, agent settings, and worktree handling got smoother for daily agent work.
- Plugins under .claude/skills are auto-loaded, so private workflow extensions no longer need marketplace packaging first.
- claude plugin init <name> scaffolds local plugins directly; useful for turning repeated repo workflows into reusable skills.
- settings.json now honors the agent field for dispatched sessions, with --agent <name> available for one-off overrides.
- EnterWorktree can switch between Claude-managed worktrees mid-session, reducing context breaks during parallel agent tasks.
- Several fixes target terminals, image paste, resume state, and background agents; less time lost to CLI session cleanup.
Source: github.com/anthropics/claude-code/releases/tag/v2.1.157Read original →
FIG-8281:1
80radar
FIG-8281:1

Fri, May 291 dispatches

#0827
#0827Agents & tools Claude Code Releases2 days ago
`Claude Code` `v2.1.154`: `Opus 4.8`, dynamic workflows, cheaper fast mode
90radar
Claude CodeCoding agent CLI — automates code edits and runs in terminal
Background workflows can now fan work across tens to hundreds of agents. This is a serious step toward larger autonomous coding runs, with cost and safety controls worth testing now.
- Opus 4.8 is now available with high effort by default, plus /effort xhigh for harder tasks. Expect better results on deep refactors and debugging.
- Dynamic workflows let Claude create and run multi-agent workflows in the background. Use /workflows to inspect runs instead of manually coordinating subtasks.
- Fast mode on Opus 4.8 costs 2x standard rate for 2.5x speed. Good for batch edits where latency beats token efficiency.
- /simplify now performs cleanup-only review and applies fixes. It is better suited to reuse, simplification, efficiency, and code altitude, not bug hunting.
- Safety and ops got sharper: rm -rf $HOME trailing-slash blocking, pending MCP approval display, and stronger data-exfiltration detection.
Source: github.com/anthropics/claude-code/releases/tag/v2.1.154Read original →
FIG-8271:1
90radar
FIG-8271:1

Wed, May 271 dispatches

#0826
#0826Agents & tools Claude Code Releases4 days ago
`Claude Code` `v2.1.152` adds review auto-fix and live skill reloads
80radar
Claude CodeCoding agent CLI — automates edits and reviews in terminal
Review findings can now patch the working tree directly. Skill iteration is faster too, with reload hooks and tool bans making agent workflows easier to govern.
- /code-review --fix now applies findings after review, covering reuse, simplification, and efficiency suggestions.
- /simplify now calls /code-review --fix, turning cleanup into a more automated refactor path.
- Skills and slash commands can set disallowed-tools, a practical guardrail for risky or narrow agent tasks.
- /reload-skills and reloadSkills: true let newly installed skills work in the same session, no restart needed.
- Fallback behavior improved: missing primary models switch to --fallback-model for the rest of the session.
Source: github.com/anthropics/claude-code/releases/tag/v2.1.152Read original →
FIG-8261:1
80radar
FIG-8261:1

Sat, May 231 dispatches

#0825
#0825Agents & tools Claude Code Releaseslast week
`Claude Code` `v2.1.149` fixes sandbox and PowerShell permission gaps
80radar
Claude CodeTerminal coding agent — automates code work with Claude
This is a security-heavy patch, not just polish. Workspace isolation, approval rules, and large-tree find behavior got fixes worth upgrading for.
- /usage now breaks limits usage down by skills, subagents, plugins, and each MCP server, making token and tool-cost debugging less blind.
- A PowerShell cd bypass let later commands read outside the workspace; this release closes that workspace escape path.
- Git worktree write allowlists now cover only shared .git, with hooks/ and config denied. Safer for repos using linked worktrees.
- Bash find no longer exhausts macOS file/vnode tables on large trees, reducing host-crash risk during broad repo scans.
- Smaller UX fixes hit /diff, task-list Markdown rendering, prompt history, transcript tailing, and slash-command hints.
Source: github.com/anthropics/claude-code/releases/tag/v2.1.149Read original →
FIG-8251:1
80radar
FIG-8251:1

`Claude Code` `v2.1.158` adds Auto mode on Bedrock, Vertex, and Foundry

`Claude Code` `v2.1.157` auto-loads local skill plugins

`Claude Code` `v2.1.154`: `Opus 4.8`, dynamic workflows, cheaper fast mode

`Claude Code` `v2.1.152` adds review auto-fix and live skill reloads

`Claude Code` `v2.1.149` fixes sandbox and PowerShell permission gaps